CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

nex-mcp-client-delivery

Secure, multi-tenant MCP server delivery kit for deploying remote LLM tools via Cloudflare Tunnels. Use when: Deploy a secure remote MCP bridge to a Raspberry Pi behind Cloudflare; Enforce tenant isolation so clients only see their own private data; Automate systemd service creation and sandboxing for MCP tools.

Quality

74%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./skills/nex-mcp-client-delivery/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

48%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body reads as a product brochure rather than an executable skill: it is padded with marketing and FAQ content, inconsistently bilingual (English scaffolding around a long Dutch worked example), and its concrete commands and code all depend on scripts and modules explicitly not included in the folder. The deployment workflow it describes is well-sequenced with real validation checkpoints, but nothing in it can actually be run from this bundle.

Suggestions

Remove the FAQ "Questions" section and the marketing sections ("Why this beats prompting it yourself", the Agensi/Nex AI banner and footer) — they restate existing content and consume tokens without adding instructional value.

Either ship the referenced files (setup.sh, deploy.sh, tools_client.py, auth/tools modules) in the bundle or inline the actual content needed to perform the task; as written, every command and code example depends on artifacts the skill admits are absent.

Unify the language (the worked example, rollout, and TODO sections are in Dutch while the rest is English) and move the full worked example to a one-level-deep reference file, keeping SKILL.md as a concise overview.

DimensionReasoningScore

Conciseness

Several padded sections add no instructional value: the promo banner ("Free skill by Nex AI... public guide from Agensi"), the marketing section "Why this beats prompting it yourself", a six-question FAQ ("## Questions") that largely restates earlier sections, and a closing advertisement. This is noticeably verbose per anchor 2 rather than merely "could be tightened" (anchor 3), though it never sinks to explaining basic concepts Claude already knows (anchor 1).

2 / 5

Actionability

Concrete guidance exists — "sudo ./setup.sh acme-corp 8848", "./deploy.sh acme-corp ./server", a full `@tool`-decorated Python example, and "auth.issue_token(\"zaakvoerder@acme.com\", \"acme-corp\", [\"read\"])" — but the skill itself states "Scripts, extension source, and templates from the original package are not in this folder", and the example imports modules (`auth`, `tools`) absent from the bundle, so nothing is executable as delivered. That is incomplete-but-concrete (anchor 3), not the minor gaps of anchor 4 nor the high-level-hints-only level of anchor 2.

3 / 5

Workflow Clarity

"How it works" gives a clear 4-step sequence (define tools → configure env → provision host → deploy and verify), and validation checkpoints are explicit: "/healthz checks", and on failure deploy.sh "print[s] de laatste veertig regels uit het journaal plus de terugdraairegel" with "Er blijft nooit een half uitgerolde versie draaien". It misses anchor 5 because the checkpoints are descriptions of absent scripts' behavior rather than a validate-fix-retry loop Claude can actually run, and "Nog te doen" step 1 (tunnel ingress placement) is terse.

4 / 5

Progressive Disclosure

Section headers organize the document, but there are no bundle files at all — every referenced file (`setup.sh`, `deploy.sh`, `tools_client.py`, `.env`, `README-voor-de-klant.md`) is missing per the skill's own disclaimer — so navigation to detail material is impossible. Large blocks that belong in separate files (the full Dutch worked example with tables, code, and test output; the FAQ) are inlined, matching anchor 3's "could be better organized" rather than anchor 4's appropriately-placed content.

3 / 5

Total

12

/

20

Passed

Description

87%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong description: concrete, technology-specific actions with an explicit multi-clause "Use when:" trigger list and a well-delineated niche. Minor improvements possible by adding the audit-logging/scoping capabilities and a few more natural synonyms (e.g., "Model Context Protocol", client IDE names).

DimensionReasoningScore

Specificity

The description lists several concrete actions — "deploying remote LLM tools via Cloudflare Tunnels", "Deploy a secure remote MCP bridge to a Raspberry Pi behind Cloudflare", "Automate systemd service creation and sandboxing for MCP tools" — grounded in specific technologies. It falls short of 5 because capabilities the body emphasizes (audit logging, granular tool scoping, health checks) are absent, and "Secure, multi-tenant MCP server delivery kit" carries mild buzzword padding; it is clearly above 3, which covers only 1-2 concrete actions.

4 / 5

Completeness

It explicitly answers both parts: the what ("Secure, multi-tenant MCP server delivery kit for deploying remote LLM tools via Cloudflare Tunnels") and an explicit, concrete when ("Use when: Deploy a secure remote MCP bridge to a Raspberry Pi behind Cloudflare; Enforce tenant isolation...; Automate systemd service creation..."). This matches the anchor-5 pattern of a clear what plus explicit multi-clause trigger phrases, exceeding anchor 4 where the when is only loosely specified.

5 / 5

Trigger Term Quality

Natural terms a user would say are present: "MCP server", "Cloudflare", "Raspberry Pi", "tenant isolation", "systemd", "remote MCP bridge". Coverage is good but not comprehensive — synonyms and variations like "Model Context Protocol" spelled out, "Claude Desktop", "Cursor", or "expose/publish MCP endpoint" are missing, so it sits between anchor 4 and the exhaustive synonym coverage of anchor 5.

4 / 5

Distinctiveness Conflict Risk

The niche is sharply defined — remote, multi-tenant MCP delivery behind Cloudflare Tunnels on a Raspberry Pi/edge device — with trigger phrases that include distinguishing technologies, so it would rarely fire for generic local-stdio MCP skills or unrelated deployment skills. Anti-drift check: anchor 4 ("minor overlap risk with closely related skills") is too weak a fit given how specific the triggers are.

5 / 5

Total

18

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository

NexaiGuy/nex-ai-skills

Path

skills/nex-mcp-client-delivery/SKILL.md

Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.